Model comparison
GLM-5 vs Qwen3.5 397B
Head-to-head evidence from 38 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.
Public leaderboard positions: GLM-5 #28 (Supported); Qwen3.5 397B #71 (Estimated). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.
Evidence parity. GLM-5 and Qwen3.5 397B share 38 comparable benchmark results. 7 of 8 categories are comparable. 11 results are unique to GLM-5; 17 to Qwen3.5 397B.
Updated July 22, 2026- Shared results
- 38
- GLM-5 only
- 11
- Qwen3.5 397B only
- 17
- Comparable categories
- 7 / 8
Pick GLM-5 if you want the stronger benchmark profile. Qwen3.5 397B only becomes the better choice if mathematics is the priority.
Confidence note. This is a partial-evidence comparison with 38 shared benchmark results across 7 evidence categories; 7 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.
Why this result
GLM-5 is clearly ahead on the BenchAlign aggregate, 66.06 to 57.01. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.
GLM-5's sharpest advantage is in knowledge, where it averages 66.4 against 56.6. The single biggest benchmark swing on the page is HLE, 50.4% to 28.7%. Qwen3.5 397B does hit back in mathematics, so the answer changes if that is the part of the workload you care about most.
Qwen3.5 397B is also the more expensive model on tokens at $0.60 input / $3.60 output per 1M tokens, versus $1.00 input / $3.20 output per 1M tokens for GLM-5. GLM-5 gives you the larger context window at 200K, compared with 128K for Qwen3.5 397B.
Category breakdown
Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.
| Category | GLM-5 | Δ | Qwen3.5 397B |
|---|---|---|---|
| Math | GLM-556.3 | Margin→ 34.3 | Qwen3.5 397B90.6 |
| Knowledge | GLM-566.4 | Margin← 9.8 | Qwen3.5 397B56.6 |
| Reasoning | GLM-560.8 | Margin→ 2.4 | Qwen3.5 397B63.2 |
| Multilingual | GLM-583.1 | Margin→ 1.6 | Qwen3.5 397B84.7 |
| Agentic | GLM-556.2 | Margin→ 0.3 | Qwen3.5 397B56.5 |
| Coding | GLM-566.3 | Margin→ 0.2 | Qwen3.5 397B66.5 |
| Inst. Following | GLM-592.6 | MarginTie | Qwen3.5 397B92.6 |
| Multimodal | GLM-5Not measured | MarginNo overlap | Qwen3.5 397B79.6 |
Decisive benchmark drivers
The largest measured benchmark gaps in this matchup, with exact reported values.
More
- Source ↗
HLE
KnowledgeA 50.4%B 28.7%Winner: GLM-5Δ 21.7HLE: GLM-5 scored 50.4%; Qwen3.5 397B scored 28.7%. GLM-5 wins this benchmark. - Source ↗
SWE-bench Pro
CodingA 55.1%B 50.9%Winner: GLM-5Δ 4.2SWE-bench Pro: GLM-5 scored 55.1%; Qwen3.5 397B scored 50.9%. GLM-5 wins this benchmark. - Source ↗
Terminal-Bench 2.0
AgenticA 56.2%B 52.5%Winner: GLM-5Δ 3.7Terminal-Bench 2.0: GLM-5 scored 56.2%; Qwen3.5 397B scored 52.5%. GLM-5 wins this benchmark. - Source ↗
SuperGPQA
KnowledgeA 66.8%B 70.4%Winner: Qwen3.5 397BΔ 3.6SuperGPQA: GLM-5 scored 66.8%; Qwen3.5 397B scored 70.4%. Qwen3.5 397B wins this benchmark. - Source ↗
AIME26
MathA 95.8%B 93.3%Winner: GLM-5Δ 2.5AIME26: GLM-5 scored 95.8%; Qwen3.5 397B scored 93.3%. GLM-5 wins this benchmark.
Operational comparison
Runtime and commercial metrics are compared only when both models have a complete sourced value.
| Metric | GLM-5 | Qwen3.5 397B | Comparison |
|---|---|---|---|
| Input / output priceUSD per 1M tokens | GLM-5$1 input / $3.2 output | Qwen3.5 397B$0.6 input / $3.6 output | Listed prices are equal. |
| Generation speedtokens per second | GLM-574 tok/s | Qwen3.5 397B96 tok/s | Qwen3.5 397B has the higher measured throughput. |
| First-answer latencyseconds to first token | GLM-51.64 s | Qwen3.5 397B2.44 s | GLM-5 reaches the first token sooner. |
| Context windowmaximum listed tokens | GLM-5200K | Qwen3.5 397B128K | GLM-5 lists the larger context window. |
Benchmark Deep Dive
AgenticQwen3.5 397B wins19 benchmarks
| Benchmark | GLM-5 | Qwen3.5 397B | Result |
|---|---|---|---|
| Terminal-Bench 2.0Source | 56.2% | 52.5% | GLM-5 leads |
| Claw-EvalSource | 57.7% | 56.8% | GLM-5 leads |
| QwenClawBenchSource | 54.1% | 51.8% | GLM-5 leads |
| τ³-bench resultsSource | 65.6% | 68.4% | Qwen3.5 397B leads |
| DeepPlanningSource | 14.6% | 37.6% | Qwen3.5 397B leads |
| ToolathlonSource | 38% | 36.3% | GLM-5 leads |
| MCP AtlasSource | 31.1% | 46.1% | Qwen3.5 397B leads |
| MCP-TasksSource | 60.8% | 74.2% | Qwen3.5 397B leads |
| WideResearchSource | 69.8% | 74.0% | Qwen3.5 397B leads |
| τ²-bench resultsSource | 98.2% | 95.6% | GLM-5 leads |
| CyberGymSource | 43.2% | — | Not comparable |
| APEX-Agents-AASource | 14.5% | 15.3% | Qwen3.5 397B leads |
| Gert LabsSource | 50.99% | 46.76% | GLM-5 leads |
| BrowseCompSource | — | 62% | Not comparable |
| VITA-BenchSource | — | 43.7% | Not comparable |
| ResearchClawBenchSource | — | 14.2% | Not comparable |
| AA Agentic IndexSource | — | 19.9% | Not comparable |
| GDPval-AASource | — | 23.1% | Not comparable |
| GDPval-AASource | — | 962 | Not comparable |
CodingQwen3.5 397B wins9 benchmarks
| Benchmark | GLM-5 | Qwen3.5 397B | Result |
|---|---|---|---|
| SWE-bench VerifiedSource | 77.8% | 76.2% | GLM-5 leads |
| SWE-bench Verified*Source | 72.8% | — | Not comparable |
| SWE-bench ProSource | 55.1% | 50.9% | GLM-5 leads |
| SWE MultilingualSource | 73.3% | — | Not comparable |
| SWE-RebenchSource | 62.8% | — | Not comparable |
| React Native EvalsSource | 74.8% | — | Not comparable |
| AA-SciCodeSource | 46.2% | 42.0% | GLM-5 leads |
| LiveCodeBench v6Source | — | 83.6% | Not comparable |
| AA Coding IndexSource | — | 48.2% | Not comparable |
ReasoningQwen3.5 397B wins4 benchmarks
KnowledgeGLM-5 wins14 benchmarks
| Benchmark | GLM-5 | Qwen3.5 397B | Result |
|---|---|---|---|
| GPQASource | 86% | 88.4% | Qwen3.5 397B leads |
| GPQA-DSource | 86.0% | — | Not comparable |
| SuperGPQASource | 66.8% | 70.4% | Qwen3.5 397B leads |
| MMLU-ProSource | 85.7% | 87.8% | Qwen3.5 397B leads |
| MMLU-Pro (Arcee)Source | 85.8% | — | Not comparable |
| HLESource | 50.4% | 28.7% | GLM-5 leads |
| Artificial Analysis Intelligence IndexSource | 39.5% | 33.7% | GLM-5 leads |
| AA-GPQA DiamondSource | 82.0% | 89.3% | Qwen3.5 397B leads |
| AA-HLESource | 27.2% | 27.3% | Qwen3.5 397B leads |
| AA-Omniscience IndexSource | 2.0% | -29.8% | GLM-5 leads |
| AA-Omniscience AccuracySource | 26.9% | 31.4% | Qwen3.5 397B leads |
| AA-Omniscience Hallucination RateSource | 34.0% | 89.1% | GLM-5 leads |
| MMLU-ReduxSource | — | 94.9% | Not comparable |
| C-EvalSource | — | 93% | Not comparable |
MathQwen3.5 397B wins8 benchmarks
| Benchmark | GLM-5 | Qwen3.5 397B | Result |
|---|---|---|---|
| AIME26Source | 95.8% | 93.3% | GLM-5 leads |
| AIME25 (Arcee)Source | 93.3% | — | Not comparable |
| HMMT Feb 2025Source | 97.5% | 94.8% | GLM-5 leads |
| HMMT Nov 2025Source | 96.9% | 92.7% | GLM-5 leads |
| HMMT Feb 2026Source | 86.4% | 87.9% | Qwen3.5 397B leads |
| MMAnswerBenchSource | 82.5% | 80.9% | GLM-5 leads |
| FrontierMath v2 (Tiers 1-3)Source | 16.434% | — | Not comparable |
| FrontierMath v2 (Tier 4)Source | 2.100% | — | Not comparable |
MultilingualQwen3.5 397B wins2 benchmarks
Multimodal8 benchmarks
| Benchmark | GLM-5 | Qwen3.5 397B | Result |
|---|---|---|---|
| Design Arena WebsiteSource | 1278 | — | Not comparable |
| MMMU-ProSource | — | 79% | Not comparable |
| MathVisionSource | — | 88.6% | Not comparable |
| CharXivSource | — | 80.8% | Not comparable |
| VideoMMMUSource | — | 84.7% | Not comparable |
| ScreenSpot ProSource | — | 65.6% | Not comparable |
| V*Source | — | 95.8% | Not comparable |
| AA-MMMU-ProSource | — | 77.3% | Not comparable |
Frequently Asked Questions (8)
Which is better, GLM-5 or Qwen3.5 397B?
GLM-5 is ahead on BenchLM's BenchAlign leaderboard, 66.06 to 57.01. The biggest single separator in this matchup is HLE, where the scores are 50.4% and 28.7%.
Which is better for knowledge tasks, GLM-5 or Qwen3.5 397B?
GLM-5 has the edge for knowledge tasks in this comparison, averaging 66.4 versus 56.6. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.
Which is better for coding, GLM-5 or Qwen3.5 397B?
Qwen3.5 397B has the edge for coding in this comparison, averaging 66.5 versus 66.3. Inside this category, SWE-bench Pro is the benchmark that creates the most daylight between them.
Which is better for math, GLM-5 or Qwen3.5 397B?
Qwen3.5 397B has the edge for math in this comparison, averaging 90.6 versus 56.3. Inside this category, HMMT Nov 2025 is the benchmark that creates the most daylight between them.
Which is better for reasoning, GLM-5 or Qwen3.5 397B?
Qwen3.5 397B has the edge for reasoning in this comparison, averaging 63.2 versus 60.8. Inside this category, AI-Needle is the benchmark that creates the most daylight between them.
Which is better for agentic tasks, GLM-5 or Qwen3.5 397B?
Qwen3.5 397B has the edge for agentic tasks in this comparison, averaging 56.5 versus 56.2. Inside this category, DeepPlanning is the benchmark that creates the most daylight between them.
Which is better for instruction following, GLM-5 or Qwen3.5 397B?
GLM-5 and Qwen3.5 397B are effectively tied for instruction following here, both landing at 92.6 on average.
Which is better for multilingual tasks, GLM-5 or Qwen3.5 397B?
Qwen3.5 397B has the edge for multilingual tasks in this comparison, averaging 84.7 versus 83.1. Inside this category, NOVA-63 is the benchmark that creates the most daylight between them.
Related Comparisons
Explore More
Choose a model with this week’s evidence
Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.
One email each week. Unsubscribe anytime.